[jira] Commented: (HADOOP-4038) Ability for JobTracker to drain running jobs before a shutdown

["Pete Wyckoff (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/HADOOP-4038?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12626225#action_12626225 ] 

Pete Wyckoff commented on HADOOP-4038:
--------------------------------------

should we just describe this as ability for admin to stop the JT from accepting new jobs and re-enable accepting new jobs?? would be nice if the caller is able to detect this and handle appropriately - i.e., the JobClient can pause until the JT is ready.

also need an API getAcceptingJobsState();




> Ability for JobTracker to drain running jobs before a shutdown
> --------------------------------------------------------------
>
>                 Key: HADOOP-4038
>                 URL: https://issues.apache.org/jira/browse/HADOOP-4038
>             Project: Hadoop Core
>          Issue Type: Improvement
>          Components: mapred
>            Reporter: dhruba borthakur
>
> It would be nice to have operating controls on the JobTracker that lets new jobs to get rejected while continuing to process existing jobs. This is handy while shutting down the JobTracker.
> Similarly, while restarting the JobTracker, it woudl be nice if it can go into this state where it refuses to accept new jobs. It allows the system administrator to ensure that all relevant TaskTrackers have joined the cluster before the cluster is opened for general use.

-- 
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.